Edge Device Image Storage via Dynamic Benchmarking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The rapid growth of edge devices is expected to overwhelm storage resources due to the vast volumes of data generated by IoT devices, necessitating efficient strategies to store only materially significant data while minimizing false positives and negatives.
Innovation Solution
A method and system for storing images and data on edge computing devices that involve extracting features and metadata from sensor data, comparing them to a pre-defined tolerance range and exogenous variables, and deciding whether to store the data locally or on a second device, with the ability to update the tolerance range to minimize false positives and negatives, using region-based convolutional neural networks or bidirectional generative adversarial networks for feature extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all sensor data from IoT devices is stored on edge devices, then data availability and response time are improved, but storage resources are overwhelmed
Solution Approach 1:
The patent extracts only the most significant features and metadata from sensor data using machine learning models, storing only these extracted representations rather than the complete raw data. This selective extraction allows the system to maintain data availability for critical information while dramatically reducing storage requirements on edge devices.
Solution Approach 2:
The system transforms raw sensor data into different parameter representations (features and metadata) that capture the essential information in a compressed form. By changing the data parameters from raw sensor readings to extracted features, the system achieves efficient storage while preserving the most valuable data characteristics.
2Quantity of substance
If data is filtered before storage to reduce volume, then storage efficiency is improved, but false positives and false negatives increase
Solution Approach 1:
The patent implements feedback mechanisms where the system continuously monitors and evaluates the performance of its filtering and selection processes. By using feedback from actual data patterns and outcomes, the system can adjust its feature extraction and selection criteria to minimize false positives and negatives while maintaining storage efficiency.
Solution Approach 2:
The system dynamically adjusts the parameters and thresholds of its feature extraction and selection processes based on learned patterns from the data. By changing the decision parameters for what constitutes significant data, the system can optimize the balance between filtering effectiveness and accuracy, reducing false classifications.
3Quantity of substance
If dynamic benchmarking is implemented to optimize storage decisions, then storage efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational workload by dividing the data processing pipeline into distinct stages: initial feature extraction using lightweight models, benchmarking comparisons, and selective storage decisions. This segmentation allows each component to be optimized independently, reducing overall computational complexity while maintaining storage optimization benefits.
Solution Approach 2:
The system implements dynamic benchmarking where the criteria for data storage are not fixed but adapt based on learned patterns and changing data characteristics. By making the benchmarking process dynamic rather than static, the system can optimize storage decisions efficiently without requiring overly complex predetermined rules, as the benchmarks evolve to match actual data patterns.
Data Source
AI summary
A method, computer program product, and computer system are provided for storing images and/or other data on an edge computing device in an edge computing environment. Data is received from one or more sensors operatively coupled to the edge computing device. One or more features and metadata are extracted from the received data. The extracted features and metadata are compared to a pre-defined tolerance range and one or more exogenous variables. The received data is caused to be stored on either the edge computing device or a second edge computing device based on a determination that the compared features and metadata falls within the tolerance range or outside of the tolerance range, respectively.